
import os
import numpy as np
import pandas as pd
from tqdm import tqdm
import matplotlib.pyplot as plt

import seaborn as sns
sns.set_context('notebook', font_scale=1.5, rc={'lines.markeredgewidth': 2})

C:\Users\marinag\Anaconda3\envs\visual_behavior_sdk\lib\site-packages\numpy\_distributor_init.py:32: UserWarning: loaded more than 1 DLL from .libs:
C:\Users\marinag\Anaconda3\envs\visual_behavior_sdk\lib\site-packages\numpy\.libs\libopenblas.GK7GX5KEQ4F6UYO3P26ULGBQYHGQO7J4.gfortran-win_amd64.dll
C:\Users\marinag\Anaconda3\envs\visual_behavior_sdk\lib\site-packages\numpy\.libs\libopenblas.PYQHXLVVQ7VESDPUVUADXEVJOBGHJPAY.gfortran-win_amd64.dll
  stacklevel=1)


%load_ext autoreload
%autoreload 2

%matplotlib inline


from visual_behavior.data_access import loading as loading
from allensdk.brain_observatory.behavior.behavior_project_cache import VisualBehaviorOphysProjectCache


import visual_behavior.utilities as vbu
import visual_behavior.visualization.utils as utils
import visual_behavior.data_access.utilities as utilities

import visual_behavior.visualization.behavior

import visual_behavior.visualization.ophys.platform_paper_figures as ppf
import visual_behavior.visualization.ophys.summary_figures as sf
import visual_behavior.visualization.ophys.experiment_summary_figures as df


experiments_table = loading.get_platform_paper_experiment_table()
print(len(experiments_table))

1249


cache_dir = loading.get_platform_analysis_cache_dir()
cache = VisualBehaviorOphysProjectCache.from_s3_cache(cache_dir=cache_dir)
# behavior_sessions = cache.get_behavior_session_table()


behavior_sessions = loading.get_platform_paper_behavior_session_table(include_4x2_data=False)

4787 behavior sessions in original behavior_sessions table
4787 behavior sessions after merging with project codes
4779 sessions after removing NaN session types
3667 sessions after removing VisualBehaviorMultiscope4areasx2d and Ai94
3667 sessions after adding session number, passive flag, and cell type columns
3667 sessions after adding experience_level column
3667 sessions after adding training_stage columns


behavior_sessions.project_code.unique()

array(['VisualBehaviorMultiscope', 'VisualBehaviorTask1B',
       'VisualBehavior'], dtype=object)


# make fake ophys_container_id column so the below functions will work
behavior_sessions['ophys_container_id'] = behavior_sessions.mouse_id.values
# add last familiar and first novel columns
behavior_sessions = utilities.add_date_string(behavior_sessions)
behavior_sessions = utilities.add_n_relative_to_first_novel_column(behavior_sessions)
behavior_sessions = utilities.add_first_novel_column(behavior_sessions)
behavior_sessions = utilities.add_last_familiar_active_column(behavior_sessions)

original_behavior_sessions = behavior_sessions.copy()


save_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\visual_behavior\platform_paper_plots\figure_1\behavior_metrics'


def get_save_folder(method, engaged_only, per_image):  
    folder = method
    if engaged_only: 
        folder = folder+'_engaged_only'
    if per_image:
        folder = folder+'_per_image' 
    return folder


def get_behavior_stats_for_sessions(behavior_session_ids, behavior_sessions, 
                                    method='stimulus_based', engaged_only=True, per_image=False):
    behavior_stats = pd.DataFrame()
    problem_sessions = []
    for behavior_session_id in tqdm(behavior_session_ids):
        try:
            print('loading for', behavior_session_id)
            stats = vbu.get_cached_behavior_stats(behavior_session_id, engaged_only=engaged_only, method=method, per_image=per_image)
            behavior_stats = pd.concat([behavior_stats, stats])
        except:
            print('cant load stats for', behavior_session_id)
            problem_sessions.append(behavior_session_id)

    cache_dir = vbu.get_behavior_stats_cache_dir(method=method, engaged_only=engaged_only, per_image=per_image)
    filepath = os.path.join(cache_dir, 'all_sessions.h5')
    if not os.path.exists(filepath):
        behavior_stats.to_hdf(filepath, key='data')

    # merge stats with session metadata
    behavior_stats = behavior_stats.merge(behavior_sessions, on='behavior_session_id')

    # remove 4x2 sessions
    behavior_stats = behavior_stats[behavior_stats.project_code!='VisualBehaviorMultiscope4areasx2d']
    
    return behavior_stats, problem_sessions


%%time 

method = 'stimulus_based'
engaged_only = True
per_image = False

folder = get_save_folder(method, engaged_only, per_image)

behavior_session_ids = behavior_sessions.index.values


behavior_stats, problem_sessions = get_behavior_stats_for_sessions(behavior_session_ids, behavior_sessions, 
                                    method=method, engaged_only=engaged_only, per_image=per_image)


original_behavior_stats = behavior_stats.copy()

  0%|                                                                                 | 2/3667 [00:00<03:55, 15.55it/s]

loading for 742008131
loading for 742797917
loading for 743041166
loading for 743660302

  0%|▏                                                                                | 6/3667 [00:00<04:10, 14.63it/s]

loading for 744374168
loading for 745054849
loading for 745355000

  0%|▏                                                                                | 8/3667 [00:00<04:29, 13.57it/s]

loading for 745636848
loading for 746222622
loading for 746928360

  0%|▎                                                                               | 12/3667 [00:00<04:38, 13.12it/s]

loading for 750415261
loading for 750781358
loading for 752002559

  0%|▎                                                                               | 14/3667 [00:01<04:31, 13.46it/s]

loading for 753011296
loading for 753936007
loading for 754509820


len(problem_sessions)

8


# # turn problem session list to a df
# tmp = pd.DataFrame(problem_sessions, columns=['behavior_session_id'])
# # load existing df of sessions
# save_dir = loading.get_platform_analysis_cache_dir()
# # problems_df = pd.read_csv(os.path.join(save_dir, 'problem_behavior_sessions.csv'))
# # # merge them and re-save
# # df = pd.concat([tmp, problems_df])
# df.to_csv(os.path.join(save_dir, 'problem_behavior_sessions.csv'))


# dataset = cache.get_behavior_session(818825644)


metrics = behavior_stats.keys()
metrics = metrics[1:18]
metrics

Index(['response_latency_count', 'response_latency_mean',
       'response_latency_std', 'response_latency_min', 'response_latency_25%',
       'response_latency_50%', 'response_latency_75%', 'response_latency_max',
       'hit_rate', 'fa_rate', 'number_of_engaged_go_trials',
       'number_of_engaged_hits', 'number_of_engaged_catch_trials',
       'number_of_engaged_false_alarms', 'fraction_engaged',
       'dprime_trial_corrected', 'dprime_non_trial_corrected'],
      dtype='object')


order = np.sort(behavior_stats.training_stage.unique())
training_stages = np.hstack((order[-6:], order[:6]))
training_stages

array(['TRAINING_0', 'TRAINING_1', 'TRAINING_2', 'TRAINING_3',
       'TRAINING_4', 'TRAINING_5', 'OPHYS_0', 'OPHYS_1', 'OPHYS_2',
       'OPHYS_3', 'OPHYS_4', 'OPHYS_5'], dtype=object)


color_map = utils.get_session_type_color_map()

keys = []
for training_stage in training_stages:
    keys.append([key for key in list(color_map.keys())[:-6] if training_stage in key][0])
    
colors = [list(color_map[key]) for key in keys]


c = []
for i in range(len(colors)):
    c.append([int(color) for color in colors[i]])


# metric = 'dprime_trial_corrected'

data = behavior_stats[behavior_stats.last_day_of_stage==True]

for metric in metrics: 
    figsize=(6,4)
    fig, ax = plt.subplots(figsize=figsize)
    ax = sns.boxplot(data=data, x='training_stage', y=metric, order=training_stages, palette='gray', ax=ax)
    ax.set_xticklabels(training_stages, rotation=90);

    utils.save_figure(fig, figsize, save_dir, folder, metric+'boxplot_by_training_stage')


data = behavior_stats[behavior_stats.last_day_of_stage==True]

for metric in metrics:
    figsize=(6,4)
    fig, ax = plt.subplots(figsize=figsize)
    order = np.sort(data['training_stage'].unique())
    ax = sns.pointplot(data=data, x='training_stage', y=metric, order=training_stages, 
                       join=False, ax=ax)
    ax.set_xticklabels(training_stages, rotation=90);
    
    utils.save_figure(fig, figsize, save_dir, folder, metric+'_pointplot_by_training_stage')


%%time 

method = 'stimulus_based'
engaged_only = True
per_image = True

folder = get_save_folder(method, engaged_only, per_image)

behavior_session_ids = behavior_sessions.index.values


behavior_stats, problem_sessions = get_behavior_stats_for_sessions(behavior_session_ids, behavior_sessions, 
                                    method=method, engaged_only=engaged_only, per_image=per_image)


original_behavior_stats = behavior_stats.copy()

  0%|                                                                                         | 0/3667 [00:00<?, ?it/s]

loading for 742008131

  0%|                                                                                 | 3/3667 [00:00<08:01,  7.61it/s]

loading for 742797917
loading for 743041166
loading for 743660302
loading for 744374168

  0%|▏                                                                                | 9/3667 [00:00<03:45, 16.20it/s]

loading for 745054849
loading for 745355000
loading for 745636848
loading for 746222622
loading for 746928360

  0%|▎                                                                               | 12/3667 [00:00<03:16, 18.62it/s]

loading for 750415261
loading for 750781358
loading for 752002559
loading for 753011296
loading for 753936007

  0%|▍                                                                               | 18/3667 [00:01<03:08, 19.36it/s]

loading for 754509820
loading for 754896244
loading for 755551879
loading for 756106020


behavior_stats.keys()

Index(['behavior_session_id', 'image_name', 'hit_rate', 'fa_rate',
       'response_latency_count', 'response_latency_mean',
       'response_latency_std', 'response_latency_min', 'response_latency_25%',
       'response_latency_50%', 'response_latency_75%', 'response_latency_max',
       'number_of_engaged_go_trials', 'number_of_engaged_hits',
       'number_of_engaged_catch_trials', 'number_of_engaged_false_alarms',
       'fraction_engaged', 'dprime_trial_corrected',
       'dprime_non_trial_corrected', 'equipment_name', 'full_genotype',
       'mouse_id', 'reporter_line', 'driver_line', 'sex', 'age_in_days',
       'cre_line', 'indicator', 'session_number',
       'prior_exposures_to_session_type', 'prior_exposures_to_image_set',
       'prior_exposures_to_omissions', 'ophys_session_id',
       'ophys_experiment_id', 'ophys_container_id', 'date_of_acquisition',
       'session_type', 'file_id', 'project_code', 'passive', 'cell_type',
       'experience_level', 'training_stage', 'first_day_of_stage',
       'last_day_of_stage', 'date', 'n_relative_to_first_novel', 'first_novel',
       'last_familiar_active'],
      dtype='object')


stats = behavior_stats[(behavior_stats.last_familiar_active==True)|(behavior_stats.first_novel)]


metrics = behavior_stats.keys()[2:19]
metrics

Index(['hit_rate', 'fa_rate', 'response_latency_count',
       'response_latency_mean', 'response_latency_std', 'response_latency_min',
       'response_latency_25%', 'response_latency_50%', 'response_latency_75%',
       'response_latency_max', 'number_of_engaged_go_trials',
       'number_of_engaged_hits', 'number_of_engaged_catch_trials',
       'number_of_engaged_false_alarms', 'fraction_engaged',
       'dprime_trial_corrected', 'dprime_non_trial_corrected'],
      dtype='object')


order = np.sort(stats.experience_level.unique())

colors = utils.get_colors_for_session_numbers()
colors = [colors[0], colors[4]]


data = stats.copy()

for metric in metrics:
    figsize=(6,4)
    fig, ax = plt.subplots(figsize=figsize)
    ax = sns.pointplot(data=data, x='experience_level', y=metric, order=order, 
                       hue='experience_level', hue_order=order, colors=colors, join=False, ax=ax)
    ax.set_xticklabels(order, rotation=90);
#     ax.set_ylim(0, 1)
    
    utils.save_figure(fig, figsize, save_dir, folder, metric+'_pointplot_by_experience_level')


stats = behavior_stats[(behavior_stats.last_familiar_active==True)|(behavior_stats.first_novel)]


def plot_metric_data_per_image(data, metric='dprime_trial_corrected', title='', ax=None):
    if ax is None: 
        fig, ax = plt.subplots()
    image_order = data.groupby(['image_name']).mean()[[metric]].sort_values(by=[metric]).index.values
    ax = sns.pointplot(data=data, x='image_name', y=metric, order=image_order, join=False, ax=ax)
    ax.set_title(title)
    ax.set_xticklabels(image_order, rotation=90)
    ax.set_xlabel('')
    ax.set_ylabel(metric)
#     ax.set_ylim(0,1)
    return ax


metric = 'response_latency_mean'

fig, ax = plt.subplots(2,2, figsize=(9,8))
ax = ax.ravel()

# A-B mice
data = stats[stats.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[0] = plot_metric_data_per_image(familiar_data, metric=metric, title='Familiar images\nimage set A', ax=ax[0])
# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[1] = plot_metric_data_per_image(novel_data, metric=metric, title='Novel images\nimage set B', ax=ax[1])


# B-A mice
data = stats[stats.project_code.isin(['VisualBehaviorTask1B'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[2] = plot_metric_data_per_image(familiar_data, metric=metric, title='Familiar images\nimage set B', ax=ax[2])

# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[3] = plot_metric_data_per_image(novel_data, metric=metric, title='Novel images\nimage set A', ax=ax[3])

for i in range(4):
    ax[i].set_ylim(0.35,0.55)
fig.tight_layout()


metric = 'dprime_trial_corrected'

fig, ax = plt.subplots(2,2, figsize=(9,8))
ax = ax.ravel()

# A-B mice
data = stats[stats.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[0] = plot_metric_data_per_image(familiar_data, metric=metric, title='Familiar images\nimage set A', ax=ax[0])
# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[1] = plot_metric_data_per_image(novel_data, metric=metric, title='Novel images\nimage set B', ax=ax[1])


# B-A mice
data = stats[stats.project_code.isin(['VisualBehaviorTask1B'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[2] = plot_metric_data_per_image(familiar_data, metric=metric, title='Familiar images\nimage set B', ax=ax[2])

# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[3] = plot_metric_data_per_image(novel_data, metric=metric, title='Novel images\nimage set A', ax=ax[3])

for i in range(4):
    ax[i].set_ylim(0,2)
fig.tight_layout()


def plot_hit_fa_rates(data, title='', ax=None):
    if ax is None: 
        fig, ax = plt.subplots()
    image_order = data.groupby(['image_name']).mean()[['hit_rate']].sort_values(by=['hit_rate']).index.values
    # hits
    ax = sns.pointplot(data=data, x='image_name', y='hit_rate', order=image_order, color='g', join=False, ax=ax)
    # FA
    ax = sns.pointplot(data=data, x='image_name', y='fa_rate', order=image_order, color='r', join=False, ax=ax)
    ax.set_title(title)
#     ax.set_xticklabels([image[2:] if image!='omitted' else image for image in image_order], rotation=90)
    ax.set_xticklabels(image_order, rotation=90)
    ax.set_xlabel('')
    ax.set_ylabel('response probability')
    ax.set_ylim(0,1)
    return ax


fig, ax = plt.subplots(2,2, figsize=(9,8))
ax = ax.ravel()

# A-B mice
data = stats[stats.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[0] = plot_hit_fa_rates(familiar_data, title='Familiar images\nimage set A', ax=ax[0])
# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[1] = plot_hit_fa_rates(novel_data, title='Novel images\nimage set B', ax=ax[1])


# B-A mice
data = stats[stats.project_code.isin(['VisualBehaviorTask1B'])]
# familiar data
familiar_data = data[data.experience_level=='Familiar']
ax[2] = plot_hit_fa_rates(familiar_data, title='Familiar images\nimage set B', ax=ax[2])

# novel data
novel_data = data[data.experience_level=='Novel 1']
ax[3] = plot_hit_fa_rates(novel_data, title='Novel images\nimage set A', ax=ax[3])

fig.tight_layout()


import visual_behavior.visualization.behavior as behavior


method = 'response_probability'
engaged_only = False
per_image = False

folder = get_save_folder(method, engaged_only, per_image)


familiar_sessions = behavior_sessions[behavior_sessions.last_familiar_active==True].index.values
novel_sessions = behavior_sessions[behavior_sessions.first_novel==True].index.values
print(len(familiar_sessions), len(novel_sessions))

78 78


# get response probability dataframe
familiar_response_probability = behavior.aggregate_response_probability_across_sessions(familiar_sessions, engaged_only=engaged_only)
novel_response_probability = behavior.aggregate_response_probability_across_sessions(novel_sessions, engaged_only=engaged_only)

# add metadata
familiar_response_probability = familiar_response_probability.merge(behavior_sessions, on='behavior_session_id')
novel_response_probability = novel_response_probability.merge(behavior_sessions, on='behavior_session_id')


figsize = (10,10)
fig, ax = plt.subplots(2,2, figsize=figsize)
ax = ax.ravel()

# A-B mice
familiar_data = familiar_response_probability[familiar_response_probability.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
novel_data = novel_response_probability[novel_response_probability.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
# get matrices
familiar_response_matrix = behavior.average_response_probability_across_sessions(familiar_data, sort=True)
novel_response_matrix = behavior.average_response_probability_across_sessions(novel_data, sort=True)

ax[0] = sns.heatmap(familiar_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[0])
ax[0].set_xlabel('change image name')
ax[0].set_ylabel('initial image name')
ax[0].set_title('Familiar images')

ax[1] = sns.heatmap(novel_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[1])
ax[1].set_xlabel('change image name')
ax[1].set_ylabel('initial image name')
ax[1].set_title('Novel images')


# B-A mice
familiar_data = familiar_response_probability[familiar_response_probability.project_code.isin(['VisualBehaviorTask1B'])]
novel_data = novel_response_probability[novel_response_probability.project_code.isin(['VisualBehaviorTask1B'])]
# get matrices
familiar_response_matrix = behavior.average_response_probability_across_sessions(familiar_data, sort=True)
novel_response_matrix = behavior.average_response_probability_across_sessions(novel_data, sort=True)

ax[2] = sns.heatmap(familiar_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[2])
ax[2].set_xlabel('change image name')
ax[2].set_ylabel('initial image name')
ax[2].set_title('Familiar images')

ax[3] = sns.heatmap(novel_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[3])
ax[3].set_xlabel('change image name')
ax[3].set_ylabel('initial image name')
ax[3].set_title('Novel images')

fig.tight_layout()
utils.save_figure(fig, figsize, save_dir, folder, 'response_probability_familiar_novel')


method = 'response_probability'
engaged_only = True
per_image = False

folder = get_save_folder(method, engaged_only, per_image)


familiar_sessions = behavior_sessions[behavior_sessions.last_familiar_active==True].index.values
novel_sessions = behavior_sessions[behavior_sessions.first_novel==True].index.values
print(len(familiar_sessions), len(novel_sessions))

78 78


# get response probability dataframe
familiar_response_probability = behavior.aggregate_response_probability_across_sessions(familiar_sessions, engaged_only=engaged_only)
novel_response_probability = behavior.aggregate_response_probability_across_sessions(novel_sessions, engaged_only=engaged_only)

# add metadata
familiar_response_probability = familiar_response_probability.merge(behavior_sessions, on='behavior_session_id')
novel_response_probability = novel_response_probability.merge(behavior_sessions, on='behavior_session_id')


figsize = (10,10)
fig, ax = plt.subplots(2,2, figsize=figsize)
ax = ax.ravel()

# A-B mice
familiar_data = familiar_response_probability[familiar_response_probability.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
novel_data = novel_response_probability[novel_response_probability.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope'])]
# get matrices
familiar_response_matrix = behavior.average_response_probability_across_sessions(familiar_data, sort=True)
novel_response_matrix = behavior.average_response_probability_across_sessions(novel_data, sort=True)

ax[0] = sns.heatmap(familiar_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[0])
ax[0].set_xlabel('change image name')
ax[0].set_ylabel('initial image name')
ax[0].set_title('Familiar images\nimage set A')

ax[1] = sns.heatmap(novel_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[1])
ax[1].set_xlabel('change image name')
ax[1].set_ylabel('initial image name')
ax[1].set_title('Novel images\nimage set B')


# B-A mice
familiar_data = familiar_response_probability[familiar_response_probability.project_code.isin(['VisualBehaviorTask1B'])]
novel_data = novel_response_probability[novel_response_probability.project_code.isin(['VisualBehaviorTask1B'])]
# get matrices
familiar_response_matrix = behavior.average_response_probability_across_sessions(familiar_data, sort=True)
novel_response_matrix = behavior.average_response_probability_across_sessions(novel_data, sort=True)

ax[2] = sns.heatmap(familiar_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[2])
ax[2].set_xlabel('change image name')
ax[2].set_ylabel('initial image name')
ax[2].set_title('Familiar images\nimage set B')

ax[3] = sns.heatmap(novel_response_matrix, cmap='magma', vmin=0, vmax=1, square=True, 
                 cbar_kws={'label':'response probability', 'shrink':0.7}, ax=ax[3])
ax[3].set_xlabel('change image name')
ax[3].set_ylabel('initial image name')
ax[3].set_title('Novel images\nimage set A')

fig.tight_layout()
utils.save_figure(fig, figsize, save_dir, folder, 'response_probability_familiar_novel_engaged_only')


familiar_sessions = behavior_sessions[(behavior_sessions.last_familiar_active==True) &
                                     (behavior_sessions.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope']))].index.values
novel_sessions = behavior_sessions[(behavior_sessions.first_novel==True) & 
                                  (behavior_sessions.project_code.isin(['VisualBehavior', 'VisualBehaviorMultiscope']))].index.values
print(len(familiar_sessions), len(novel_sessions))

52 52


familiar_data = behavior.aggregate_response_probability_across_sessions(familiar_sessions, engaged_only=engaged_only)
novel_data = behavior.aggregate_response_probability_across_sessions(novel_sessions, engaged_only=engaged_only)


len(familiar_data.image_name.unique())

8


def add_image_pair_column(data):
    data = data.reset_index(drop=True)
    data['image_pair'] = [data.iloc[row]['previous_image_name'][2:]+'_'+data.iloc[row]['image_name'][2:] for row in range(len(data))]
    return data


familiar_data = add_image_pair_column(familiar_data)
novel_data = add_image_pair_column(novel_data)


familiar_pair_order = familiar_data.groupby('image_pair').mean()[['response_probability']].sort_values(by='response_probability').index.values
novel_pair_order = novel_data.groupby('image_pair').mean()[['response_probability']].sort_values(by='response_probability').index.values


figsize = (16,4)
fig, ax = plt.subplots(1, 2, figsize=figsize)

ax[0] = sns.pointplot(data=familiar_data, x='image_pair', y='response_probability', order=familiar_pair_order, 
                      color='gray', join=False, ax=ax[0])
ax[0].set_xticklabels(familiar_pair_order, rotation=90, fontsize=8)
ax[0].set_xlabel('image pair')
ax[0].set_ylim(0, 1)
ax[0].set_title('Familiar images')


ax[1] = sns.pointplot(data=novel_data, x='image_pair', y='response_probability', order=novel_pair_order, 
                      color='gray', join=False, ax=ax[1])
ax[1].set_xticklabels(novel_pair_order, rotation=90, fontsize=8)
ax[1].set_xlabel('image pair')
ax[1].set_ylim(0, 1)
ax[1].set_title('Novel images')

fig.tight_layout()


def add_image_pair_index(data, pair_order):
    data['image_pair_index'] = None
    for row in range(len(data)):
        image_pair = data.iloc[row].image_pair
        pairs = list(pair_order)
        index = pairs.index(image_pair)
        data.at[row, 'image_pair_index'] = index
    return data


familiar_data = add_image_pair_index(familiar_data, familiar_pair_order)
novel_data = add_image_pair_index(novel_data, novel_pair_order)


order = np.sort(stats.experience_level.unique())

colors = utils.get_colors_for_session_numbers()
colors = [colors[0], colors[3]]


order = np.sort(familiar_data.image_pair_index.unique())

figsize = (8,4)
fig, ax = plt.subplots(figsize=figsize)

ax = sns.pointplot(data=familiar_data, x='image_pair_index', y='response_probability', order=order, 
                      color=colors[0], label='Familiar', join=False, ax=ax)
ax = sns.pointplot(data=novel_data, x='image_pair_index', y='response_probability', order=order, 
                      color=colors[1], label='Novel', join=False, ax=ax)
ax.set_xticklabels(order, rotation=90, fontsize=8)
ax.set_xlabel('image pair (sorted)')
ax.set_ylim(0, 1)
ax.legend(['Familiar - image set A', 'Novel - image set B'], loc='upper left', fontsize='small')
ax.set_title('behavior performance across image transitions\n VisualBehavior & VisualBehaviorMultiscope')
ax.set_ylabel('response probability')

Text(0, 0.5, 'response probability')


low_perf_images = familiar_data[familiar_data.image_pair_index.isin(np.arange(8,32))].image_name.unique()


familiar_experiment_id = experiments_table[experiments_table.behavior_session_id==familiar_data.behavior_session_id.values[0]].index.values[0]
dataset = loading.get_ophys_dataset(familiar_experiment_id)

C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))


stimulus_templates = dataset.stimulus_templates.copy()
fig, ax = plt.subplots(2,4,figsize=(10,5))
ax=ax.ravel()
for i,image_name in enumerate(stimulus_templates.index):
    ax[i].imshow(stimulus_templates.loc[image_name]['warped'], cmap='gray')
    if image_name in low_perf_images:
        color='r'
    else:
        color='k'
    ax[i].set_title(image_name, color=color)
    ax[i].axis('off')


novel_low_perf_images = novel_data[novel_data.image_pair_index.isin(np.arange(8,32))].image_name.unique()


novel_experiment_id = experiments_table[experiments_table.behavior_session_id==novel_data.behavior_session_id.values[0]].index.values[0]
dataset = loading.get_ophys_dataset(novel_experiment_id)

C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '795953296' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '795953296' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "


stimulus_templates = dataset.stimulus_templates.copy()
fig, ax = plt.subplots(2,4,figsize=(10,5))
ax=ax.ravel()
for i,image_name in enumerate(stimulus_templates.index):
    ax[i].imshow(stimulus_templates.loc[image_name]['warped'], cmap='gray')
    if image_name in novel_low_perf_images:
        color='r'
    else:
        color='k'
    ax[i].set_title(image_name, color=color)
    ax[i].axis('off')


novel_low_perf_images

array(['im073', 'im106', 'im075', 'im031', 'im035', 'im000', 'im054'],
      dtype=object)


# does it matter what image it changed from?
novel_data[novel_data.image_pair_index.isin(np.arange(8,32))].previous_image_name.unique()

array(['im106', 'im075', 'im031', 'im035', 'im000', 'im054', 'im045'],
      dtype=object)


np.sort(novel_data[novel_data.image_pair_index.isin(np.arange(8,32))].image_pair.unique())

array(['000_031', '000_035', '000_073', '000_075', '031_035', '031_054',
       '031_073', '031_075', '031_106', '035_073', '035_075', '035_106',
       '045_035', '045_054', '045_073', '045_075', '054_000', '054_031',
       '054_035', '054_073', '054_075', '075_035', '075_073', '106_073'],
      dtype=object)


familiar_sessions = behavior_sessions[(behavior_sessions.last_familiar_active==True) &
                                     (behavior_sessions.project_code.isin(['VisualBehaviorTask1B']))].index.values
novel_sessions = behavior_sessions[(behavior_sessions.first_novel==True) & 
                                  (behavior_sessions.project_code.isin(['VisualBehaviorTask1B']))].index.values
print(len(familiar_sessions), len(novel_sessions))

familiar_data = behavior.aggregate_response_probability_across_sessions(familiar_sessions, engaged_only=engaged_only)
novel_data = behavior.aggregate_response_probability_across_sessions(novel_sessions, engaged_only=engaged_only)

familiar_data = add_image_pair_column(familiar_data)
novel_data = add_image_pair_column(novel_data)

familiar_pair_order = familiar_data.groupby('image_pair').mean()[['response_probability']].sort_values(by='response_probability').index.values
novel_pair_order = novel_data.groupby('image_pair').mean()[['response_probability']].sort_values(by='response_probability').index.values

familiar_data = add_image_pair_index(familiar_data, familiar_pair_order)
novel_data = add_image_pair_index(novel_data, novel_pair_order)

26 26


colors = utils.get_colors_for_session_numbers()
colors = [colors[0], colors[3]]
sns.palplot(colors)

order = np.sort(familiar_data.image_pair_index.unique())

figsize = (8,4)
fig, ax = plt.subplots(figsize=figsize)

ax = sns.pointplot(data=familiar_data, x='image_pair_index', y='response_probability', order=order, 
                      color=colors[0], label='Familiar - image set B', join=False, ax=ax)
ax = sns.pointplot(data=novel_data, x='image_pair_index', y='response_probability', order=order, 
                      color=colors[1], label='Novel - image set A', join=False, ax=ax)
ax.set_xticklabels(order, rotation=90, fontsize=8)
ax.set_xlabel('image pair')
ax.set_ylim(0, 1)
ax.legend(['Familiar - image set B', 'Novel - image set A'], loc='upper left', fontsize='small')
ax.set_title('behavior performance across image transitions\n VisualBehaviorTask1B')
ax.set_ylabel('response probability')

Text(0, 0.5, 'response probability')


low_perf_images = novel_data[novel_data.image_pair_index.isin(np.arange(8,22))].image_name.unique()

novel_experiment_id = experiments_table[experiments_table.behavior_session_id==novel_data.behavior_session_id.values[0]].index.values[0]
dataset = loading.get_ophys_dataset(familiar_experiment_id)

stimulus_templates = dataset.stimulus_templates.copy()
fig, ax = plt.subplots(2,4,figsize=(10,5))
ax=ax.ravel()
for i,image_name in enumerate(stimulus_templates.index):
    ax[i].imshow(stimulus_templates.loc[image_name]['warped'], cmap='gray')
    if image_name in low_perf_images:
        color='r'
    else:
        color='k'
    ax[i].set_title(image_name, color=color)
    ax[i].axis('off')

C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))


mouse_id = behavior_sessions.mouse_id.unique()[0]
mouse_data = behavior_sessions[behavior_sessions.mouse_id==mouse_id]


training_stages = np.sort(mouse_data[mouse_data.training_stage.str.contains('TRAINING')].training_stage.unique())


behavior_session_id = mouse_data[(mouse_data.training_stage==training_stages[0])&(mouse_data.first_day_of_stage)].index.values[0]
dataset = cache.get_behavior_session(behavior_session_id, skip_eye_tracking=True)

C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))


import visual_behavior.visualization.ophys.experiment_summary_figures as ep


figsize = (10,5)
fig, ax = plt.subplots(1,2, figsize=figsize)
trials = dataset.trials.copy()
trials = trials.reset_index()
trials = trials[(trials.auto_rewarded==False)&(trials.aborted==False)]
go_trials = trials[trials.go==True]
catch_trials = trials[trials.catch==True]

ax[0] = ep.plot_lick_raster(go_trials, response_window=dataset.task_parameters['response_window_sec'], ax=ax[0], save_dir=None)
ax[1] = ep.plot_lick_raster(catch_trials, response_window=dataset.task_parameters['response_window_sec'], ax=ax[1], save_dir=None)

c:\users\marinag\documents\code\visual_behavior_analysis\visual_behavior\visualization\ophys\experiment_summary_figures.py:111: UserWarning: Attempting to set identical bottom == top == 0 results in singular transformations; automatically expanding.
  ax.set_ylim(0, len(trials))
c:\users\marinag\documents\code\visual_behavior_analysis\visual_behavior\visualization\ophys\experiment_summary_figures.py:111: UserWarning: Attempting to set identical bottom == top == 0 results in singular transformations; automatically expanding.
  ax.set_ylim(0, len(trials))


behavior_session_ids = []
for i, training_stage in enumerate(training_stages):
    behavior_session_id = mouse_data[(mouse_data.training_stage==training_stage)&(mouse_data.first_day_of_stage)].index.values[0]
    behavior_session_ids.append(behavior_session_id)
    behavior_session_id = mouse_data[(mouse_data.training_stage==training_stage)&(mouse_data.last_day_of_stage)].index.values[0]
    behavior_session_ids.append(behavior_session_id)
    
data_dict = {}
for behavior_session_id in behavior_session_ids:
    dataset = cache.get_behavior_session(behavior_session_id)
    data_dict[behavior_session_id] = dataset

C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '742008131' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '742008131' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '742008131' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '742008131' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '742797917' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '742797917' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '745355000' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '745355000' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '745636848' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '745636848' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '746222622' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '746222622' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '746928360' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '746928360' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '750781358' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '750781358' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '752002559' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '752002559' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\AppData\Roaming\Python\Python37\site-packages\hdmf\spec\namespace.py:533: UserWarning: Ignoring cached namespace 'hdmf-common' version 1.3.0 because version 1.5.0 is already loaded.
  % (ns['name'], ns['version'], self.__namespaces.get(ns['name'])['version']))
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\rig_geometry.py:135: UserWarning: This nwb file with identifier '772658979' has no eye tracking rig metadata. (NWB error: "'eye_tracking_rig_metadata' not found in processing of NWBFile 'root'.")
  warnings.warn("This nwb file with identifier "
C:\Users\marinag\Documents\Code\AllenSDK\allensdk\brain_observatory\behavior\data_objects\eye_tracking\eye_tracking_table.py:136: UserWarning: This nwb file with identifier '772658979' has no eye tracking data. (NWB error: 'EyeTracking')
  warnings.warn("This nwb file with identifier "


n_stages = len(training_stages)

figsize = (20,10)
fig, ax = plt.subplots(2, n_stages, figsize=figsize)
ax = ax.ravel()

for i, training_stage in enumerate(training_stages):
    behavior_session_id = mouse_data[(mouse_data.training_stage==training_stage)&(mouse_data.first_day_of_stage)].index.values[0]
    dataset = data_dict[behavior_session_id]
    trials = dataset.trials.copy()
    trials = trials.reset_index()
    trials = trials[(trials.auto_rewarded==False)&(trials.aborted==False)]

    ax[i] = ep.plot_lick_raster(trials, response_window=dataset.task_parameters['response_window_sec'], ax=ax[i], save_dir=None)
    ax[i].set_title(training_stage+', day 1')
    
    
    behavior_session_id = mouse_data[(mouse_data.training_stage==training_stage)&(mouse_data.last_day_of_stage)].index.values[0]
    dataset = data_dict[behavior_session_id]
    trials = dataset.trials.copy()
    trials = trials.reset_index()
    trials = trials[(trials.auto_rewarded==False)&(trials.aborted==False)]

    ax[i+n_stages] = ep.plot_lick_raster(trials, response_window=dataset.task_parameters['response_window_sec'], ax=ax[i+n_stages], save_dir=None)
    ax[i+n_stages].set_title(training_stage+', day -1')
fig.tight_layout()

c:\users\marinag\documents\code\visual_behavior_analysis\visual_behavior\visualization\ophys\experiment_summary_figures.py:111: UserWarning: Attempting to set identical bottom == top == 0 results in singular transformations; automatically expanding.
  ax.set_ylim(0, len(trials))
c:\users\marinag\documents\code\visual_behavior_analysis\visual_behavior\visualization\ophys\experiment_summary_figures.py:111: UserWarning: Attempting to set identical bottom == top == 0 results in singular transformations; automatically expanding.
  ax.set_ylim(0, len(trials))


%%time 

method = 'stimulus_based'
engaged_only = True
per_image = False

folder = get_save_folder(method, engaged_only, per_image)

behavior_session_ids = behavior_sessions.index.values


behavior_stats, problem_sessions = get_behavior_stats_for_sessions(behavior_session_ids, behavior_sessions, 
                                    method=method, engaged_only=engaged_only, per_image=per_image)


original_behavior_stats = behavior_stats.copy()

  0%|                                                                                 | 5/3667 [00:00<01:13, 49.93it/s]

loading for 742008131
loading for 742797917
loading for 743041166
loading for 743660302
loading for 744374168
loading for 745054849
loading for 745355000
loading for 745636848
loading for 746222622
loading for 746928360
loading for 750415261

  0%|▎                                                                               | 16/3667 [00:00<01:16, 47.54it/s]

loading for 750781358
loading for 752002559
loading for 753011296
loading for 753936007
loading for 754509820
loading for 754896244
loading for 755551879
loading for 756106020
loading for 756632727

  1%|▌                                                                               | 27/3667 [00:00<01:21, 44.89it/s]

loading for 757431083
loading for 758162988
loading for 758942142
loading for 759251920
loading for 759756829
loading for 760007428
loading for 760409951
loading for 760698062
loading for 761499069
loading for 761831761
loading for 762178061

  1%|▊                                                                               | 37/3667 [00:00<01:20, 45.16it/s]

loading for 762722108
loading for 763291103
loading for 763980798
loading for 764495320
loading for 764897570
loading for 765396393
loading for 766110635
loading for 766753111
loading for 767275980
loading for 768042187

  1%|█                                                                               | 47/3667 [00:01<01:22, 43.96it/s]

loading for 768815421
loading for 769452637
loading for 769902997
loading for 770244427
loading for 771245588
loading for 772125590
loading for 772658979
loading for 773585522
loading for 774106776


order = np.sort(behavior_stats.training_stage.unique())
training_stages = np.hstack((order[-6:], order[:6]))
training_stages

array(['TRAINING_0', 'TRAINING_1', 'TRAINING_2', 'TRAINING_3',
       'TRAINING_4', 'TRAINING_5', 'OPHYS_0', 'OPHYS_1', 'OPHYS_2',
       'OPHYS_3', 'OPHYS_4', 'OPHYS_5'], dtype=object)


# useful info
# "novel training" is the first day of images after gratings
experience_levels = ['Gratings', 'Novel Training', 'Familiar Training', 'Familiar', 'Novel 1', 'Novel >1']
print(experience_levels)
cell_types = np.sort(behavior_sessions.cell_type.unique())

['Gratings', 'Novel Training', 'Familiar Training', 'Familiar', 'Novel 1', 'Novel >1']


metric = 'dprime_trial_corrected'
# colors = utils.get_experience_level_colors()
colors = utils.get_cre_line_colors()

fig, ax = plt.subplots(figsize=(10,5))
ax = sns.boxplot(data=behavior_stats, x='experience_level', y=metric, order=experience_levels, 
                  hue='cell_type', hue_order=cell_types, palette=colors)
ax.legend(fontsize='small', title='', bbox_to_anchor=(1,1))
ax.set_xticklabels(experience_levels, rotation=45)
ax.set_xlabel('')

fig.tight_layout()


metric = 'dprime_trial_corrected'
colors = utils.get_experience_level_colors()
colors = [(0.5, 0.5, 0.5)]+colors

exp_levels = ['Training', 'Familiar', 'Novel 1', 'Novel >1']
data = behavior_stats[behavior_stats.experience_level.isin(exp_levels)]

fig, ax = plt.subplots(figsize=(8,4))
ax = sns.boxplot(data=data, x='cell_type', y=metric, order=cell_types, 
                  hue='experience_level', hue_order=exp_levels, palette=colors, width=0.5, dodge=True)
ax.legend(fontsize='small', title='', bbox_to_anchor=(1,1))
ax.set_xlabel('')
ax.set_ylabel('d-prime')
fig.tight_layout()


metric = 'dprime_trial_corrected'
colors = utils.get_experience_level_colors()
colors = [(0.5, 0.5, 0.5)]+colors

metrics = ['dprime_non_trial_corrected',
       'dprime_trial_corrected', 'hit_rate', 'fa_rate', 'fraction_engaged',
       'number_of_engaged_catch_trials',
       'number_of_engaged_false_alarms', 'number_of_engaged_go_trials',
       'response_latency_mean', 'response_latency_std',]

exp_levels = ['Training', 'Familiar', 'Novel 1', 'Novel >1']
data = behavior_stats[behavior_stats.experience_level.isin(exp_levels)]

for metric in metrics: 
    fig, ax = plt.subplots(figsize=(8,4))
    ax = sns.boxplot(data=data, x='cell_type', y=metric, order=cell_types, 
                      hue='experience_level', hue_order=exp_levels, palette=colors, width=0.5, dodge=True)
    ax.legend(fontsize='small', title='', bbox_to_anchor=(1,1))
    ax.set_xlabel('')
    fig.tight_layout()

get experiments and behavior sessions tables¶

add useful columns for filtering¶

save directory for plots¶

load behavior metrics¶

plot behavior metrics across stages¶

get behavior metrics for each image (instead of for all go/catch trials as above)¶

plot metrics for last familiar and first novel sessions (average across all images)¶

plot behavior metrics for each image for familiar & novel image sets¶

hit and FA rates per image¶

plot response probability heatmaps for all image transitions in last familiar and first novel sessions¶

full session¶

engaged portions of sessions only¶

sort and plot performance across all image transitions for A-B mice¶

whats up with the images in the range where familiar is lower than novel?¶

which novel images are in the same range in the ordered set?¶

repeat for B-A mice¶

is it the same set of images that are harder to detect for image set A (Novel set in this case)?¶

plot lick raster for first and last stage of each session type for one mouse¶

get behavior session object for sessions of interest¶

Plot metrics distributions (averaged across images within each session)¶

plot stuff by cre line¶

color by exp level per cre line¶